Hazards of Forecasting

January 9, 1956

“Cromwell was about to ravage all Christendom; the royal family was undone, and his own forever established, save for a little grain of sand which formed in his ureter. Rome herself was trembling under him; but this small piece of gravel having formed there, he is dead, his family cast down, all is peaceful, and the king is restored.”

Perhaps nothing in literature more concisely and effectively illustrates the way in which apparently tiny causes can have tremendous effects than these two sentences from the great seventeenth-century mathematician and philosopher, Pascal. The world had another telling illustration of the hazards of forecasting in the year just passed. When President Eisenhower’s heart faltered on Sept. 24, the whole economic outlook seemed to change overnight. On the Monday after that week end, the stock market opened with the most violent drop since the panicky days of 1929, later to recover with the President’s own recovery. Not one of those who had forecast business and market conditions at the beginning of 1955 had thought to condition his forecast upon the health of a single man.

Yet as the year turned out, the forecasters for 1955 went wrong for the strange reason that they were not optimistic enough. An article in the National Industrial Conference Board’s December Business Record, with the witty title “What Went Right?” points out that “the consensus of forecasts for 1955 was certainly far wide of the mark.” The forecasts were only moderately optimistic. In general they guessed that 1955 might be the “second best” year in our business history—better than 1954, but not as good as 1953.

When the forecasts were made, nothing in the statistical sky seemed to hint anything better than this. “While consumer demand had held up very well indeed during 1954, there was no very good evidence that it was about to advance into a wholly new range of spending.” The assignment of cause and effect can be questioned, but the Conference Board article thinks it is “reasonably clear that, in 1955, personal consumption and related magnitudes of personal saving and personal debt were the dynamic and driving forces behind business. . . . Of all the records set during the year, the most impressive are the 7.5 million level of new car sales, and the $5 billion—$6 billion increase in installment debt outstanding. No other figures so clearly distinguish the year from all others. If the forecasters had known them a year ago, they would have needed little else to get on target.” In other words, if the forecasters had known what was going to happen, they would have known what was going to happen.

If the consensus of the forecasts could go wrong for 1955, it can just as easily go wrong for 1956—and not necessarily in the same direction. It is the weakness of most forecasters to assume that conditions are going to continue as they already are—or, rather, to assume that they are going to change in the same direction and at the same rate as in the immediate past. In more technical language, the unconscious tendency of most forecasters is to “extrapolate” the recent business curve.

Let us study the present forecasts therefore, with appropriate caution. All of us must base our decisions on some guess regarding the probable future. But we should do well to keep in mind that business forecasting can never be a science. The future does not depend on any set of past or present statistics, or any extrapolation of them, but on a million future developments which no one can foresee. The best forecast can be based only upon apparent probabilities; but “probabilities” can be falsified by events.

As the Conference Board’s article concludes. “It is not easy to say what the lesson of 1955 should be. Should the performance of consumption in 1955, which was extraordinary by the standard of all preceding years, be counted on for 1956? . . . Or does the boom in personal spending in 1955 presage . . . a steep return toward an historical ‘normal’? . . . The answers to these questions are very much the answers to 1956.”